Loki + OpenTelemetry — Prerequisites
What to Know Before Starting Loki
Loki is specifically approachable if you already know Prometheus — LogQL deliberately mirrors PromQL's structure, and Loki's label-based indexing is the same mental model as Prometheus's label-based metrics, just applied to logs instead of numbers.
Required (Must Have)
1. Prometheus/PromQL Familiarity
If you haven't been through this site's Prometheus section yet, do that first — Loki's Learning Modules assume you already understand labels, streams (Loki's equivalent of a metric series), and the rate/aggregation mental model, and build on it rather than re-teaching it.
2. Basic Grafana Navigation
Loki is Grafana's native log backend — you'll be exploring logs inside Grafana's Explore view, not a separate standalone Loki UI:
3. Kubernetes Basics (for the Promtail/Agent deployment path)
Most production Loki deployments run Promtail (or the newer Grafana Agent/Alloy) as a DaemonSet — the same deployment pattern as Datadog's Agent or Prometheus's node_exporter.
Nice to Have (Speeds Up Learning)
ELK/Filebeat Background
If you've used Elasticsearch/Logstash/Kibana before, Overview's framing directly applies: Loki is a deliberately cheaper alternative for high-volume logs specifically because it indexes only labels, not full log content — understanding what ELK does differently makes that tradeoff concrete rather than abstract.
JSON/logfmt Parsing Familiarity
LogQL's pipeline stages (| json, | logfmt) assume basic comfort reading structured log formats — you don't need to write a parser, just recognize the shapes.
YAML Comfort
Promtail's scrape configs and pipeline stages are YAML-based, following the same kubernetes_sd_configs pattern as Prometheus's own service discovery — prior Prometheus YAML exposure transfers directly.
What You Do NOT Need
Time Estimate
If you have the prerequisites above (especially Prometheus/PromQL):
Start Here
Go to the Installation section to deploy Loki and Promtail, then proceed to Fundamentals.